Neural Modeling of Parison Extrusion in Extrusion Blow Molding

Abstract

Back-propagation (BP) neural network is used to develop process models for the parison extrusion in extrusion blow molding based on experimental data. In applying the BP network, some modifications, such as using a self-adaptive learning rate coefficient, determining the number of hidden neurons through experimentation, and so on, to the original BP algorithm are carried out to speed up learning. Quite a good agreement has been reached between the predicted parison length and swells using the trained BP models and the experimentally determined ones. The prediction of the parison diameter and thickness distributions can be made online at any parison length or any parison drop time within a given range using the trained models. It has been demonstrated that nonlinear swells, under the effect of sag, can be predicted within a reasonably adequate accuracy.